A structured U.S.-PRC government dialogue on strategic artificial intelligence risk reduction - The US and China should be talking about AI safety and governance, and while I am skeptical this can happen given the politics I thought it is still constructive to make some suggestions. I created this proposal with Astra after much back and forth. If the two sides are going to dialogue about AI, the AI should have a say
Pangram informs me this newsletter is 99% AI. Please, no more slop.
I’m very eager to hear your human, expert thoughts on the policy.
I’m also alarmed at the anthropomorphism here.
Astra is not a person or being with a “say.” It is a human-created piece of software based in simple linear algebra, which is usually taught to high-school seniors or college freshman, at the latest. It is a collection of algorithms, like all other software, a 100% normal technology.
The Chinese have a firm understanding of this, due to Xi’s emphasis on AI education at all levels, which Americans do not due to a lack of education & over-reliance on hype & deliberately misleading metaphors in the media offered by Silicon Valley.
Please exercise your stellar critical thinking here, Bill. Keep up your quality work & best wishes.
Please post a more serious, studious follow-up based in your human intelligence & writing. Thanks for listening.
Did you read the opening paragraph? I said explicitly I created it after much back and forth with Astra because since the dialogue would be about ai I thought the ai should have a say. My usual newsletters are 100% human
Yes Bill, I read it. And I nearly wept. Please hear me with an open mind, because I’m speaking in good faith here as someone deeply sympathetic to you.
Substack is alreadyawash in slop on this topic, precisely when we need deep analysis.
The question of how China will go forward with the current 5 year plan on AI & how the US & the world will join or reject their actions is indeed key towards 2030, deserving the kind of expert human understanding & judgment that only you can provide.
With deepest, complete & sincere respect, that you hold that a collection of repeated matrix multiplication algorithms could “have a say,” is troubling anthropomorphism.
Your Excel spreadsheet contains built in functions for common calculations, some complex, & generates answers, yet no one believes an Excel spreadsheet “has a say.” Please consider this fact.
Both programs are just mathematical algorithms, Bill. They have no opinions, ideas, or understanding. They cannot judge, decide, or synthesize information. They do not create. They cannot hold an opinion. The LLMs are merely running a non-deterministic algorithm to pick the most likely token, while Excel is deterministic. As you know.
That we humans see fluid generated text & associate it with any understanding, “say,” or position is our projection of Theory of Mind.
And we do this because we mistakenly believe fluid language is the sign of mind & intelligence.
Truly I’m not dissing you or downing you, Bill. You’re great. Honestly. I often recommend that people follow & subscribe to you. Yesterday I suggested the famed Gary Marcus talk to you about the current 5-year plan, particularly Chapters 13 & 14.
I learn so much from you regularly, for which I’m grateful. If I didn’t think of you highly, I wouldn’t take this moment to speak to you with such candor.
Thanks for listening Bill. Have a great rest of the week. 🙏
Pangram informs me this newsletter is 99% AI. Please, no more slop.
I’m very eager to hear your human, expert thoughts on the policy.
I’m also alarmed at the anthropomorphism here.
Astra is not a person or being with a “say.” It is a human-created piece of software based in simple linear algebra, which is usually taught to high-school seniors or college freshman, at the latest. It is a collection of algorithms, like all other software, a 100% normal technology.
The Chinese have a firm understanding of this, due to Xi’s emphasis on AI education at all levels, which Americans do not due to a lack of education & over-reliance on hype & deliberately misleading metaphors in the media offered by Silicon Valley.
Please exercise your stellar critical thinking here, Bill. Keep up your quality work & best wishes.
Please post a more serious, studious follow-up based in your human intelligence & writing. Thanks for listening.
Did you read the opening paragraph? I said explicitly I created it after much back and forth with Astra because since the dialogue would be about ai I thought the ai should have a say. My usual newsletters are 100% human
Yes Bill, I read it. And I nearly wept. Please hear me with an open mind, because I’m speaking in good faith here as someone deeply sympathetic to you.
Substack is alreadyawash in slop on this topic, precisely when we need deep analysis.
The question of how China will go forward with the current 5 year plan on AI & how the US & the world will join or reject their actions is indeed key towards 2030, deserving the kind of expert human understanding & judgment that only you can provide.
With deepest, complete & sincere respect, that you hold that a collection of repeated matrix multiplication algorithms could “have a say,” is troubling anthropomorphism.
Your Excel spreadsheet contains built in functions for common calculations, some complex, & generates answers, yet no one believes an Excel spreadsheet “has a say.” Please consider this fact.
Both programs are just mathematical algorithms, Bill. They have no opinions, ideas, or understanding. They cannot judge, decide, or synthesize information. They do not create. They cannot hold an opinion. The LLMs are merely running a non-deterministic algorithm to pick the most likely token, while Excel is deterministic. As you know.
That we humans see fluid generated text & associate it with any understanding, “say,” or position is our projection of Theory of Mind.
And we do this because we mistakenly believe fluid language is the sign of mind & intelligence.
Truly I’m not dissing you or downing you, Bill. You’re great. Honestly. I often recommend that people follow & subscribe to you. Yesterday I suggested the famed Gary Marcus talk to you about the current 5-year plan, particularly Chapters 13 & 14.
I learn so much from you regularly, for which I’m grateful. If I didn’t think of you highly, I wouldn’t take this moment to speak to you with such candor.
Thanks for listening Bill. Have a great rest of the week. 🙏